Targeting Erosion Control Using Spatially Distributed Sediment Budgets

Sediment Budgets 2 (Proceedings of symposium S1 held during the Seventh IAHS Scientific Assembly at Foz do Iguaçu, Brazil, April 2005). IAHS Publ. 292, 2005. 65

Targeting erosion control using spatially distributed sediment budgets

SCOTT N. WILKINSON, JON M. OLLEY, ARTHUR M. READ & RON C. DEROSE CSIRO Land and Water, PO Box 1666, ACT 2601, Australia [email protected]

Abstract There is a growing interest in restoring rivers; however, the restoration work required often exceeds the resources available. Consequently, management actions need to be targeted to achieve the greatest benefit. A thorough knowledge of local sediment budgets could represent a substantial management technique in this regard. The SedNet model has been used extensively to construct local sediment budgets throughout river networks. In the model, sediment is supplied to each link of the network from hillslope erosion, gully erosion, and channel bank erosion, and then routed through the network, with deposition occurring in river channels, on flood plains, and in reservoirs. The model enables the determination of the contribution from each erosion source, at each point in the river network. The SedNet model is applied to the Murrumbidgee River, and shows how spatially distributed sediment budgets can be used to target erosion control activities. Key words riverbank erosion; sediment budget; sediment tracing; SedNet; targeting

INTRODUCTION

The extent and diversity of physical habitat is an important determinant of the health of river ecosystems. Physical habitat includes the size and shape of the channel, the form of the bed sediments, the quality of the water, and the supply of nutrients, all of which are influenced by the supply and characteristics of the sediments (Prosser et al., 2001a). European settlement, 100–200 years ago, brought widespread changes in land- use across the southeast of Australia, which massively increased the supply of sediment to the rivers (e.g. Olley & Wasson, 2003). Today, increased sediment supply continues to affect the health of many Australian rivers. Unfortunately, the magnitude of work required to reduce sediment loads to “natural”, pre-European settlement rates exceeds the resources available. To maximize the benefit from investment, erosion control works must, therefore, be targeted to the most important sediment sources in the river basin. In this paper, spatially distributed sediment budgets are used to assess the relative magnitude and spatial patterns of sediment sources in the Murrumbidgee Basin, in southeastern Australia. Two contrasting methods for targeting erosion control activities are discussed herein.

Basin description

The Murrumbidgee River drains approximately 84 000 km2 of the Murray-Darling Basin (Fig. 1). Its catchment has three distinct geomorphic regions: upper, middle, and

66 Scott N. Wilkinson et al.

GRIFFITH Mid Catchment Mu HAY rrumb idgee River Burrinjuck Dam

GUNDAGAI WAGGA WAGGA

CANBERRA

Lower Catchment

Blowering Dam 5025 0 50 100

Kilometres 5025 0 50 100 Kilometres Upper Catchment Fig. 1 Map of the Murrumbidgee River basin showing key locations.

lower. The upper Murrumbidgee is mountainous and hilly, and is separated from the mid-region by two large storage reservoirs, Burrinjuck and Blowering (Fig. 1), which trap most of the sediment delivered from the upper basin (Wasson et al., 1987) and regulate the flow downstream. The mid-basin has undulating terrain dissected by incised stream and gully networks. Several radionuclide sediment tracing studies have shown erosion of these networks to be the primary source of sediment delivered to the lower Murrumbidgee River (e.g. Wallbrink et al., 1998; Olley & Wasson 2003). A strong rainfall gradient exists across the basin; average annual totals vary from 1600 mm in the upper catchment down to 300 mm at the catchment outlet. The 29 000 km2 upper and middle catchment is the primary sediment source area for the basin (Olive et al., 1996), and targeting erosion control in this region only, is discussed. The main land-use in the upper and middle basin is grazing, with some cropping, forestry, and national parks.

Constructing the sediment budget

The SedNet model, developed for the National Land and Water Resources Audit (NLWRA; Prosser et al., 2001b), is a physically-based process model. It constructs sediment budgets for a river network and identifies the major sources, sinks, and loads of material. In the model, the river network is divided into a series of links which are the basic unit of calculation for the sediment budget. Each link extends between adjacent stream junctions or nodes, and has a sub-catchment that drains into the link between its upper and lower nodes (Fig. 2). ArcInfoTM AML scripts are used to define the river network and sub-catchments from a 25 m Digital Elevation Model (DEM). Using this environment, separate budgets for bedload and suspended sediment are then

Targeting erosion control using spatially distributed sediment budgets 67

computed for each link. The sediment yield (Yx ) from a link is given by: = + + + − − − Yx H x Gx Bx Tx Cx Fx Rx (1)

Sediment inputs to each link come from hillslope (Hx), gully (Gx), and riverbank (Bx) erosion, and from upstream tributaries (Tx) (Fig. 3). Sediment is either deposited within the channel (Cx), on flood plains (Fx), in reservoirs (Rx), or is transported downstream and delivered to the next link (Yx). This process is carried out in each river link, from first to highest Shreve order (Shreve, 1966), so that Y x for the outlet link, represents mean annual export.

1 3

2 1

1 1

Fig. 2 A river network showing links, nodes, Shreve magnitude of each link (Shreve, 1966) and the internal sub-catchment areas of an order one and an order four link.

Hillslope erosion (t/y) Riverbank Gully Tributary supply (t/y) erosion (t/y) erosion (t/y) HSDR

Channel deposition

Floodplain area Downstream yield (t/y)

Fig. 3 Conceptual diagram of the SedNet sediment budget for a river link.

Hillslope erosion The input from hillslope erosion is estimated using the Revised Universal Soil Loss Equation, as applied in the NLWRA (Lu et al., 2001). Most soil eroded on hillslopes deposits elsewhere on the hillslope, and so only a small amount of the eroded sediment is delivered to streams. Hence, H x is modified by the hillslope sediment delivery ratio (HSDR). A HSDR of 0.05 has been found to provide good agreement between hillslope erosion and yield in regions dominated by hillslope processes (Prosser et al., 2001b). All hillslope sediment is assumed to contribute to the suspended sediment budget.

Gully erosion The linear extent of gully erosion in the Murrumbidgee basin has been mapped from aerial photographs by the New South Wales Department of

68 Scott N. Wilkinson et al.

Infrastructure, Planning, and Natural Resources. Assuming negligible gully erosion prior to the arrival of Europeans, the average rate of suspended sediment supply since, from gullies in each sub-catchment (Gx), is the product of gully length (lx), cross- 2 -3 sectional area (a =12 m ), and sediment dry bulk soil density (ρs =1.5 t m ), divided by the average time over which gullies have developed (τ = 120 years): ρ a G = s × l (2) x τ x Sediment from gullies and riverbanks is distributed equally to separate bedload and suspended load budgets. The channel network extension that occurred in response to European land-use change is, today, largely complete. Suspended sediment supply from gullies is therefore reduced, in the model, from the long-term average by 50%, to reflect a decline in the sediment yield, indicated by a historical survey in the Murrumbidgee (Wasson et al., 1998), and recent measurements (Caitcheon, 2004).

Riverbank erosion Following DeRose et al. (this volume), bank erosion rate, -1 BE x (m year ), is determined from an empirical model as: = ρ − − − BEx 0.0001 gQbf S x (1 PRx )(1 exp( 0.008Fw )) (3) Negligible bank erosion has been assumed in the proportion of the link length that has fully intact riparian vegetation (PRx), as determined from LANDSAT imagery with 30 m pixels (Barson et al., 2000). The coefficient (0.0001) was calibrated to achieve a maximum bank erosion rate of 0.5 m year-1; in accordance with observed channel widening in some highly eroded, steep and non-vegetated streams. The value used differs from the coefficient (0.00002) used elsewhere (DeRose et al., this volume), which indicates that the bank erosion model may be relatively poorly constrained. The = ρ amount of sediment supplied from bank erosion, Bx s hLx (BEx ) , where h is bank height (3 m) and Lx is link length.

Flood plain deposition Deposition of suspended sediment on flood plains, Fx (t year-1), was determined as a function of flood plain area, and median overbank flow (Prosser et al., 2001a). Flood plain mapping for each river link utilised hydrological and hydraulic modelling, (Pickup & Marks, 2001). Bankfull flow was estimated as the flow with a recurrence interval of 2.27 years (Page, 1988). Using daily river gauge records, bankfull and median overbank flow were regionalized as power functions of mean annual flow, which was also regionalized (Wilkinson et al., 2005). Downstream of reservoirs, regionalized flow was adjusted to reflect flow regulation.

Channel deposition Where bedload supply to a link exceeds transport capacity (Prosser & Rustomji, 2000), the excess bedload (Cx) is deposited in the channel.

Deposition in reservoirs Sediment deposition in reservoirs (Rx) is predicted using an empirical function of the mean annual reservoir inflow and its total storage capacity (Heinemann, 1981).

Using sediment budgets for targeting erosion control Where the objective is to improve water quality throughout the river network, the total sediment supply to the

Targeting erosion control using spatially distributed sediment budgets 69

network should be reduced. This is most effectively achieved by targeting the erosion process(es) that supply the most sediment. Long term averages generally are sufficient for catchment planning, and maximum reduction in sediment supply is achieved by spatially targeting control measures to the sub-catchments and river-links with the highest rates of erosion. In contrast, if the objective is to reduce sediment export from a catchment, the sub-catchments contributing the greatest amount of sediment to the catchment outlet should be targeted. Methods of determining the areas contributing to export are described more fully in Wilkinson et al. (2005) and Lu et al. (2004). This study illustrates how the SedNet model can be used to locate erosion control measures to achieve these two different management objectives.

RESULTS AND DISCUSSION

The modelled stream network in the Upper and Middle Murrumbidgee catchment has a total length of 5600 km. The network is composed of 745 river links; of 6.5 km average length. The predicted long-term average suspended sediment load at Wagga Wagga is 578 kt year-1, which is similar to the observed load of 600 kt year-1 (Olive et al. 1996). Summing each term in the bedload and suspended sediment budgets for all links in the river network identifies riverbank and gully erosion as dominant sediment sources (Table 1); this agrees with radionuclide tracer observations of suspended sediment (Wallbrink et al., 1998). Therefore, erosion control to reduce the total supply of sediment, and also sediment contribution to export, should target both riverbank and gully erosion. The link-average riverbank erosion rate is reduced in links with high proportions of existing bank vegetation. To calculate the bank erosion rate along the non-vegetated portion of the link, bank erosion hazard (t km-1 -1 year ) is calculated, by setting PRx in equation (3) to zero, then dividing Bx by link length. This hazard represents the potential erosion rate, whether or not it has been realised to date. There are strong spatial patterns evident in bank erosion hazard (Fig. 4). It is highest along main streamlines and in steeper foothill areas, and lowest in the flat plains of the northwest. Spatially distributed sediment budgets can be used to help target erosion control to meet different management objectives. The predicted results from two contrasting scenarios, each involving the same management action; stabilisation of 500 km of eroding riverbank, and 500 km of gullies, are compared. In the first scenario, the objective is to reduce total sediment supply to the stream network. To achieve this

Table 1 Predicted relative proportions of sediment sources and losses in the upper-mid Murrumbidgee catchment. Inputs % of total Outputs % of total Hillslope suspended supply 19 Flood plain suspended deposition 9 Gully suspended supply 12 Channel bedload deposition 21 Gully bedload supply 24 Reservoir suspended deposition 20 Riverbank suspended supply 23 Reservoir bedload deposition 16 Riverbank bedload supply 23 Export suspended sediment 24 Export bedload sediment 9 Total supply 100% Total output 100%

70 Scott N. Wilkinson et al.

objective, the actions need to be targeted at the links with the highest bank erosion hazard and the sub-catchments with the greatest gully density (Fig. 4). The objective of the second scenario is to reduce catchment export of suspended sediment. In this case,

Fig. 4 Bank erosion hazard (t-1 km-1 year-1) and Gully erosion (t-1 ha-1 year-1).

Fig. 5 Contribution from bank erosion hazard and gully erosion, to export of suspended sediment.

Targeting erosion control using spatially distributed sediment budgets 71

Table 2 Comparison between scenarios with different objectives for targeting erosion control. Targeting Objective Reduction in total sediment supply Reduction in suspended å ()+ + sediment export (%) H x Gx Bx (%) Reduce sediment supply 18 23 Reduce suspended sediment 16 26 export

actions should be targeted to links with the highest bank erosion hazard contribution to export, and sub-catchments with the highest gully contribution to export (Fig. 5). In each treated link and sub-catchment, the scenarios simulate stabilization of 50% of the non-vegetated portion of the link, and stabilization of 50% of gully erosion. (Figs 4 and 5 indicate, in dark shade, links containing 1000 km of non-vegetated river link, and 1000 km of gully erosion). This partial treatment accounts for two pragmatic considerations. Firstly, erosion rates will vary within links and sub-catchments, and 50% erosion control may be sufficient to stabilize erosion hotspots within links and sub-catchments. Also, not all landholders will want to be involved in erosion control. The two scenarios are compared in Table 2. The different effect on suspended sediment export between the scenarios demonstrates the importance of carefully identifying the objective of targeting erosion control. The sediment-supply scenario targets bank erosion control to links with high stream power throughout the river system, whereas the contribution-to-export scenario targets bank erosion downstream of the major reservoirs. Spatially distributed sediment budgets can be modelled at a range of scales, depending on the area over which erosion control is being targeted, and the spatial resolution required. For example, the method also has been applied to the Murray- Darling Basin; of which the Murrumbidgee is a subset (Lu et al., 2004). Sediment tracing techniques (Olley, 2004), can be used to test the model predictions.

CONCLUSIONS

Spatially distributed sediment budgets are a useful modelling framework for identifying sediment sources and sinks in a river basin. The SedNet model can be used to target erosion control activities to the dominant erosion process, and to erosion hotspots, to achieve different management objectives. It also can be used to test the efficacy of different management scenarios. It is currently being used to plan erosion control in the Murrumbidgee catchment.

Acknowledgements This research was supported by grant CLW44 from Land & Water Australia. We also acknowledge Heinz Buettikofer and Tristram Miller (CSIRO Land and Water) for preparing Figs 1 and 4, respectively.

REFERENCES

Barson, M. M., Randell, L. A., & Bordas, V. (2000) Land Cover Change in Australia. Bureau of Rural Sciences, Canberra, Australia.

72 Scott N. Wilkinson et al.

Caitcheon, G. (2004) Unpublished data on sediment yield from gully networks in the Murrumbidgee catchment. CSIRO Land and Water, Canberra, Australia. Heinemann, H. G. (1981) A new sediment trap efficiency curve for small reservoirs. Water Resour. Res. 17, 825–830. Lu, H., Gallant, J., Prosser, I., Moran, C. & Priestly, G. (2001) Prediction of Sheet and Rill Erosion over the Australian Continent, Incorporating Monthly Soil Loss Distribution. Tech. Report 13/01, CSIRO Land and Water, Canberra, Australia. Lu, H., Moran, C. J., Prosser, I. P. & DeRose, R. (2004) Investment prioritization based on broadscale spatial budgeting to meet downstream targets for suspended sediment loads. Water Resour. Res. 40, W09501, doi 10.1029/2003WR002966. Olive, L. J., Olley, J. M., Wallbrink, P. J. & Murray, A. S. (1996) Downstream patterns of sediment transport during floods in the Murrumbidgee River, NSW, Australia. Z. Geomorph. N. F. Suppl.-Bd. 105, 129–140. Olley, J. M. & Wasson, R. J. (2003) Changes in the flux of sediment in the Upper Murrumbidgee catchment, SE Australia, since European settlement. Hydrol. Processes 17, 3307–3320. Page, K. J. (1988) Bankfull discharge frequency for the Murrumbidgee River, New South Wales. In: Fluvial Geomorphology of Australia, 267–281. Academic Press, Australia. Pickup, G. & Marks, A. (2001) Identification of floodplains and estimation of floodplain flow velocities for sediment transport modelling. Tech. Report 14/01, CSIRO Land and Water, Canberra, Australia. Prosser, I. P. & Rustomji, P. (2000) Sediment transport capacity relations for overland flow: Progr. in Phys. Geogr. 24(2), 179–193. Prosser, I. P., Rutherfurd I. D., Olley, J. M., Young, W. J., Wallbrink, P. J. & Moran, C. J. (2001a) Large-scale patterns of erosion and sediment transport in rivers networks, with examples from Australia. Mar. Freshwater Res. 52, 81–99. Prosser, I., Young, B., Rustomji, P., Moran, C. & Hughes, A. (2001b) Constructing river basin sediment budgets for the national land and water resources audit. Tech. Report 15/01, CSIRO Land and Water, Canberra, Australia. Rutherfurd, I. D. (2000) Some human impacts on Australian stream channel morphology. In: River Management: The Australasian experience (ed. by S. Brizga & B. Finlayson), 11–49. John Wiley and Sons, Chichester, UK. Shreve, R. L. (1966) Statistical Law of Stream Numbers. J. Geol. 74, 17–37. Wallbrink, P. J., Murray, A. S., Olley, J. M. & Olive, L. (1998) Determining sediment sources and transit times of suspended sediments in the Murrumbidgee River, NSW, Australia using fallout Cs-137 and Pb-210. Water Resour. Res. 34, 879–887. Wasson, R. J., Clark, R. L. & Nanninga, P. M. (1987) 210Pb as a chronometer and tracer, Burrinjuck Reservoir, Australia. Earth Surf. Processes Landf. 12, 399–414. Wasson, R. J., Mazari, R. K., Starr, B. & Clifton, G. (1998) The recent history of erosion and sedimentation on the Southern Tablelands of southeastern Australia. Geomorphol. 24, 291–308. Wilkinson, S. N., Prosser, I. P., Olley, J. M. & Read, A. M. (2005) Using sediment budgets to prioritise erosion control in large river systems. In: River/Catchment Dynamics: Natural Processes and Human Impacts (Proc. Int. Conf. Solsona, Spain). IAHS Publ. (submitted). IAHS Press, Wallingford, UK.